Expression-driven vs. Keyframe-based animation for high-layer-count repetitive motion

Direct Answer

There is no fixed layer-count threshold where expressions become unusable. Expressions typically preserve maintainability for hundreds of synchronized layers, but keyframes become preferable when expression evaluation prevents stable real-time preview or dominates render time. The trade-off depends on expression complexity, hardware, After Effects version, and comp settings—not layer count alone.

Confirmed Behavior

• Expressions are evaluated by the expression engine at request time (preview, render, scrub). Keyframes store precomputed values.
• Converting an expression to keyframes (Animation → Keyframe Assistant → Convert Expression to Keyframes) bakes the current result and removes runtime evaluation for that property.
• Baking is one-way unless you keep an expression-driven master comp or duplicate.

Likely Explanation

Overhead scales with the number and cost of evaluated expression properties, dependency chains, and frames requested. Simple per-layer expressions (e.g., thisComp.layer("Controller").transform.position) on hundreds of layers may be acceptable. A few heavy expressions with loops, string parsing, sampleImage(), or cross-comp lookups can be worse than thousands of simple links. Multi-frame rendering and caching behavior also vary by AE release, so the same comp can perform differently after upgrades.

Practical Verification Steps

1. Duplicate the comp. In one copy, select expression-driven properties and convert to keyframes. Keep the original.
2. Compare Composition panel preview responsiveness at target resolution and frame rate.
3. Render both copies with identical output settings and compare render durations.
4. If only preview is slow, test with draft settings, proxies, or temporarily disabling expressions before baking.
5. Re-test after AE updates or hardware changes.

Optimization Before Conversion

• Simplify expressions: avoid per-frame string operations, unnecessary property lookups, and repeated calculations.
• Reference a single controller layer or null instead of chaining expressions.
• Reduce affected properties (e.g., drive only Position, not Position+Scale+Rotation).
• Use guide layers or precomps for static/repetitive elements.
• Consider pre-rendering heavy sections to footage.

Diagnostic Question

What is the actual bottleneck—Composition panel preview lag, final render time, or both—and are the expressions simple property links or complex per-frame calculations? This determines whether baking, optimization, or preview settings changes are the right next step.